Human Diversity
Page 27
Second, parents can make a negative difference at the extremes. Really awful parenting, involving severe deprivation and abuse, can damage children permanently.[10] To raise children in really awful neighborhoods, where gangs run the streets and the schools are dangerous and chaotic, increases the chances that your children will be snared in peer groups (part of the nonshared environment) that damage their life chances.[11] Proposition #8 should be understood to refer to parenting and neighborhoods within a broad range but excluding the worst of the worst.
Third, the limited role of the family environment does not apply to life outcomes that can be directly determined by money. It applies only to cognitive repertoires. If you are wealthy you can give your children wealth, regardless of their personalities, abilities, or social behaviors. Your socioeconomic class has a causal role in determining your children’s material standard of living, but you cannot use your socioeconomic status to make your children more than trivially “better” than they would have been otherwise, whether “better” is defined in terms of personality, abilities, or social behavior.
Can you use your money to make your children professionally more successful than they would otherwise have been? It depends on the profession. The arena in which family wealth can clearly make a huge difference is politics, because wealth can influence nominating committees and finance campaigns, and the powers associated with a political office automatically go to the winner of enough votes, regardless of talent or character. If the parents own the corporation, they can bequeath the power of ownership to the next generation (e.g, the Murdochs, Sulzbergers, and Kochs). Parental wealth and influence can help a child get the appropriate degree and help obtain the first job. It’s not so easy for parental influence to get the child promoted. The more competitive the industry and the more cognitively demanding the job, the less influence family wealth has.
The Polderman Meta-Analysis of Twin Studies
The generality of Proposition #8 is most economically established with a single source, a meta-analysis of twin studies published in 2015. The study was conducted by a team of seven Dutch, Australian, and American scholars. It was conceived and led by Danielle Posthuma, head of the Department of Complex Trait Genetics at Vrije Universiteit in Amsterdam. First author was Tinca J. C. Polderman.12 It was a mammoth undertaking, effectively covering all twin studies from 1958 to 2012. The article reporting the results, “Meta-analysis of the Heritability of Human Traits Based on Fifty Years of Twin Studies,” involved 2,748 publications and 14,558,903 twin pairs that explored 17,804 traits.
The authors found that the ACE model, which is limited to additive genetic variance, is usually appropriate. If the difference in the correlations between MZ and DZ twins is solely due to additive genetic variance, then the null hypothesis is that 2rDZ = rMZ. The authors reported that the “observed pattern of twin correlations is consistent with a simple and parsimonious underlying model of the absence of environmental effects shared by twin pairs and the presence of genetic effects that are entirely due to additive genetic variation.”13
In instances when the ACE model produced an estimate of the shared environment less than zero (2rDZ – rMZ < 0), the model that had to be substituted leads to an underestimation of h2, as explained in the note.[14]
The table below shows the traits that fall into the three categories of cognitive repertoires that I have used throughout the book: characteristics of personality, abilities, and social behavior. I ordered the traits in each category from low to high in the shared environment’s estimated role.
TRAITS DIRECTLY RELATED TO COGNITIVE REPERTOIRES
Personality
Specific personality disorders
Shared environment: 1%
Nonshared environment: 56%
Genes: 44%
Temperament and personality functions
Shared environment: 5%
Nonshared environment: 44%
Genes: 44%
Emotionally unstable personality
Shared environment: 19%
Nonshared environment: 46%
Genes: 35%
Abilities
Energy and drive functions
Shared environment: 0%
Nonshared environment: 43%
Genes: 57%
Experience of self and time functions
Shared environment: 0%
Nonshared environment: 44%
Genes: 56%
Psychomotor functions
Shared environment: 1%
Nonshared environment: 69%
Genes: 30%
Attention functions
Shared environment: 2%
Nonshared environment: 55%
Genes: 44%
Memory functions
Shared environment: 3%
Nonshared environment: 52%
Genes: 45%
Calculation functions
Shared environment: 13%
Nonshared environment: 32%
Genes: 56%
Higher-level cognitive functions
Shared environment: 18%
Nonshared environment: 27%
Genes: 55%
Language functions
Shared environment: 22%
Nonshared environment: 32%
Genes: 46%
Mild mental retardation
Shared environment: 22%
Nonshared environment: 45%
Genes: 33%
Social behavior
Work and employment
Shared environment: 0%
Nonshared environment: 63%
Genes: 37%
Intimate relationships
Shared environment: 0%
Nonshared environment: 65%
Genes: 35%
Attention functions
Shared environment: 2%
Nonshared environment: 55%
Genes: 44%
Family relationships
Shared environment: 6%
Nonshared environment: 66%
Genes: 28%
Informal social relationships
Shared environment: 10%
Nonshared environment: 59%
Genes: 32%
Global psychosocial functions
Shared environment: 11%
Nonshared environment: 41%
Genes: 48%
Societal attitudes
Shared environment: 12%
Nonshared environment: 50%
Genes: 37%
Looking after one’s health
Shared environment: 13%
Nonshared environment: 43%
Genes: 44%
Conduct disorders
Shared environment: 14%
Nonshared environment: 38%
Genes: 48%
Disorders due to tobacco use
Shared environment: 17%
Nonshared environment: 29%
Genes: 54%
Recreation and leisure
Shared environment: 18%
Nonshared environment: 27%
Genes: 55%
Disorders due to alcohol use
Shared environment: 19%
Nonshared environment: 38%
Genes: 44%
Religion and spirituality
Shared environment: 21%
Nonshared environment: 43%
Genes: 36%
Disorders due to cannabinoid use
Shared environment: 22%
Nonshared environment: 25%
Genes: 54%
Educational attainment
Shared environment: 25%
Nonshared environment: 26%
Genes: 50%
Disorders due to multiple drug use
Shared environment: 26%
Nonshared environment: 29%
Genes: 46%
Problems related to upbringing
Shared environment: 34%
Nonshared environment: 40%
Genes: 27%
Basic interpersonal interactions
Shared envi
ronment: 36%
Nonshared environment: 34%
Genes: 30%
Source: Polderman, Benyamin, de Leeuw et al. (2015): Supplementary Table 21. Statistics are reported for the “best” model.
In interpreting the numbers, remember that the estimates for heritability are biased downward by about 10 percent for reasons described in the note to the table. Also, only a few of the categories discriminate by age. Thus, for example, the role of the shared environment for “higher-level cognitive functions” is 18 percent—an estimate that is too low if we are thinking about early childhood, too high if we are thinking about adults.
Of the 30 traits in the table, 11 have shared-environment roles of less than 10 percent, 11 have roles of 10–19 percent, 6 have roles of 20–26 percent, and 2 have more than 30 percent. The shared environment had its highest estimated values, 36 percent and 34 percent, for “basic personal interactions” and “problems related to upbringing.” (Yes, these data seem to say, you can have some effect on your kids’ manners and you can also cause problems.) The 11 traits with a role for the shared environment of less than 10 percent include specific personality disorders, temperament and personality functions, work and employment, intimate relationships, and family relationships.
The next table shows traits that involve serious psychological problems, all of which affect cognitive repertoires in one way or another.
PSYCHOLOGICAL DISORDERS
Anxiety and depression
Depressive episode
Shared environment: 4%
Nonshared environment: 58%
Genes: 39%
Recurrent depressive disorder
Shared environment: 4%
Nonshared environment: 44%
Genes: 52%
Other anxiety disorders
Shared environment: 9%
Nonshared environment: 49%
Genes: 42%
Phobic anxiety disorders
Shared environment: 10%
Nonshared environment: 45%
Genes: 45%
Psychological disorders associated with childhood and adolescence
Eating disorders
Shared environment: 2%
Nonshared environment: 60%
Genes: 38%
Hyperkinetic disorders
Shared environment: 5%
Nonshared environment: 27%
Genes: 68%
Pre-adult emotional and behavioral disorders
Shared environment: 7%
Nonshared environment: 29%
Genes: 64%
Emotional disorders with onset in childhood
Shared environment: 20%
Nonshared environment: 37%
Genes: 43%
Other psychological disorders
Stress and adjustment disorders
Shared environment: 0%
Nonshared environment: 67%
Genes: 33%
Nonorganic sleep disorders
Shared environment: 0%
Nonshared environment: 55%
Genes: 45%
Other adult personality and behavior disorders
Shared environment: 0%
Nonshared environment: 60%
Genes: 41%
Obsessive-compulsive disorders
Shared environment: 6%
Nonshared environment: 48%
Genes: 46%
Mood disorders
Shared environment: 6%
Nonshared environment: 32%
Genes: 63%
Pervasive developmental disorders
Shared environment: 7%
Nonshared environment: 23%
Genes: 70%
Bipolar disorder
Shared environment: 14%
Nonshared environment: 19%
Genes: 68%
Source: Polderman, Benyamin, de Leeuw et al. (2015): Supplementary Table 21. Statistics are reported for the “best” model.
Of the 15 traits in the table, only 2 have a role for the shared environment exceeding 10 percent: bipolar disorder (14 percent) and emotional disorders with onset specific to childhood (20 percent). The role of the shared environment for 12 of the 15 is less than 10 percent.
Can parents drive children to distraction? No doubt about it; just as children routinely drive parents to distraction. But when it comes to severe mental disorders, the parents’ genes are important while their parenting, by and large, is not.
Before leaving the estimates of heritability, I need to add more detail about two of the most important cognitive repertoires, the general mental factor g and personality characteristics.
The General Mental Factor g
The heritability of IQ scores (the best measures of g) was contested from the 1960s through the 1990s, but the data-driven arguments had subsided even by the time Herrnstein and I published The Bell Curve in 1994. At that time, we used a wide range for characterizing heritability, saying that it was somewhere between 40 percent and 80 percent. Since then, the state of knowledge has advanced. Here are two findings for which there was some evidence before the publication of The Bell Curve and that have been solidly established since then:
The shared environment plays a large role in determining IQ during the first few years of life, diminishing thereafter.
By the time people reach adolescence, almost all studies have found that the shared environment has a negligible relationship to IQ.
Both findings are best documented in a review of the literature by Elliot Tucker-Drob and his colleagues at the University of Texas. The graph below summarizes their results.
Source: Adapted from Tucker-Drob, Briley, and Harden (2013): Fig. 1. Shaded areas bound ± 1 standard error.
The graph combines the results of publications reporting on 11 different longitudinal twin and adoptions studies. In infancy and the first few years of life, the shared environment explains a large proportion of the variance in scores—around two-thirds. By seven years of age, that figure has dropped to about one-third. By 14, it is zero.
There is more to be said about the relative roles of heritability and the shared environment early in life, but I defer that discussion to chapter 13. For now, I confine myself to the story for IQ in adulthood. After about age 14, there is no evidence from twin and adoption studies that their shared environment as children had anything to do with their IQ scores.
Personality Characteristics
No matter whether researchers use the Big Five or one of the other personality models, the answer to the question “How much effect does the shared environment have on the way that human personalities develop?” is the same: Effectively none.
In 1987, six years after Plomin and David Rowe had named the nonshared environment, Plomin wrote the authoritative review of the state of knowledge at the time. This was his summary with regard to personality:
In a review of 10 recent twin studies of personality, the average twin correlations were .47 for identical twins and .23 for fraternal twins. This pattern of twin correlations suggests that heredity accounts for 50% of the phenotypic variance and that nonshared environment and error of measurement explain the rest.15
In other words, the value of C in the ACE model was zero. “It might seem odd,” Plomin continued, “to report average correlations across a domain as diverse as personality.” But, he added, there were hardly any exceptions. Whether the topic was a Big Five characteristic such as extraversion or neuroticism or more specific characteristics such as tolerance, sense of well-being, or alienation, twin studies of heritability kept coming up with correlations for MZ twins that were more than twice the correlations for DZ twins, leaving no role for the shared environment.16 He further noted that studies of non-twin siblings and adoption studies were consistent with those of the twin studies.17
In 2014, Turkheimer, with coauthors Erik Pettersson and Erin Horn, published a long article in the Annual Review of Psychology that reviewed the history of the study of personality. It’s worth quoting in full his summary of th
e role of the shared environment:
The Second Law of Behavior Genetics, which states that the shared environmental component of human individual differences is small, is usually true for most traits, but the situation is somewhat starker for personality. It is remarkable, in surveying the genetically informed personality literature in a very wide context, how completely absent the shared environment is. In fact, it is often the case that identical twins are more than twice as similar as fraternal twins, a violation of the classical twin model that, if uncorrected, produces negative estimates for shared environmental variance.18
Before leaving the topic of personality, a reminder is important. The shared environment includes the influences that fall under the headings of both SES and parenting. With regard to parenting, the salient finding is not that parents have no influence on their children, but that whatever influence they have is probably part of the nonshared environment. Plomin put it nicely: “From Freud onwards, theories of socialization had assumed that children’s environments are doled out on a family-by-family basis. In contrast, the point of nonshared environment is that environments are doled out on a child-by-child basis.”19
Recapitulation
The literature on shared environment, nonshared environment, and heritability tells us that a family’s SES (income, parental education and occupation) is unimportant in explaining the cognitive abilities and personality traits that parents try hardest to promote. It is a counterintuitive finding. It is an unwelcome finding for parents. But it is based on a technically strong, thoroughly replicated body of evidence.
We cannot jump from this evidence to the conclusion that SES is unimportant for explaining outcomes across families. Half a century ago, in a landmark book titled Inequality, Christopher Jencks used an example of fallacious reasoning about genes and environment that became iconic: Suppose that red-haired children are arbitrarily denied access to schooling. Their performance on academic tests suffers. Hair color is genetic. A correlation will be found between genetics and test scores, but the relationship is not causal. The causal effect of red hair is the denial of schooling, which in turn is the true cause of worse test scores.20 For assessing the relationship of SES and heritable traits, we need to turn to other bodies of evidence, the subject of the next chapter.